There has conventionally been known a method for correcting brightness variations in fluorescence images caused by observation distances and angles by dividing a fluorescence image by a reflected light image (see, for example, PTL 1 to PTL 3).